Chapter 7: Segmentation, Targeting and Positioning

Dana Okafor has a one-word problem written at the top of the whiteboard: families. Lumen Cycles plans to take the Lumen Cargo national over the next year, and the plan says the bike is "for families." Marcus Lee, the marketing analyst, has just pulled the list of everyone who bought a Cargo in Denver, Seattle, Portland and Salt Lake City, and the list doesn't look like one group of people. Some buyers ride every school day and sold a car after they bought the bike. Others take it out on a Saturday for groceries and leave it in the garage the rest of the week. Some bought it at a dealer after a long test ride with their kids in the seat; others clicked "buy" on the website after watching one video. A few don't have children at all and use the Cargo to haul gear.

"Families" is a starting point, not a target. A national campaign aimed at every parent in the country would cost more than Lumen's entire marketing budget and would say nothing in particular to anyone. Dana needs to know which families, why they would choose a $2,999 cargo e-bike over a second car or a cheaper online bike, and what Lumen should stand for in their minds. Those are the three questions of this chapter: segmentation (how is the market divided?), targeting (which parts do we serve?) and positioning (how do we want those customers to see us compared with the alternatives?).

Marketers often abbreviate the sequence as STP. It comes after the market analysis you did in earlier chapters and before the decisions about product, price, place and promotion that fill the rest of the book. If you get STP wrong, every later decision is aimed at the wrong person.

7.1 Market segmentation

Every market is made up of buyers who differ in their needs, resources, locations, attitudes and habits. A company could in principle design a separate offer for every buyer, and a company could in principle treat all buyers as the same. Most companies do neither. They look for groups of buyers who are similar to one another and different from other groups, and they decide which groups to serve.

The key words in the second definition are respond in a similar way. A segment isn't just a group of people who share a trait. Left-handed people share a trait, but left-handed people don't buy e-bikes for different reasons than right-handed people, so left-handedness is useless for segmenting the e-bike market. A segment is useful only when its members want something different, or would react differently to a price, a message or a channel, than people outside it.

There's no single right way to segment a market. Marketers try different segmentation bases, the variables used to sort buyers into groups, alone and in combination, to find the view that best reveals real differences in what people want. The four main bases for consumer markets are geographic, demographic, psychographic and behavioral.

7.1.1 Geographic segmentation

Geographic segmentation divides a market by location: countries, regions, states, cities, neighborhoods, or by features of a place such as climate, population density or terrain. It is the oldest and simplest base, and it still matters whenever needs or costs vary by place.

McDonald's keeps its core menu but adapts it country by country; its restaurants in India, for instance, serve vegetarian items built for local diets. Retailers stock snow shovels in Minneapolis in November and not in Phoenix.

For e-bikes, geography matters in several ways. Urban density affects how far people need to travel and how hard it is to park a car. Climate affects how many months a year people ride. Hills make electric assistance more valuable. And Lumen's dealer network, about 60 independent shops in the Mountain West and Pacific Northwest, means that a family in Atlanta can't currently walk into a Lumen dealer for a test ride. Geography is therefore both a description of demand and a constraint on what Lumen can offer.

7.1.2 Demographic segmentation

Demographic segmentation divides the market by variables such as age, life-cycle stage, gender, income, education, occupation, household size, ethnicity and generation. Demographic variables are popular because they are often linked to needs and usage and are easy to measure.

• Age and life-cycle stage. Needs change as people move from student to young professional to parent to empty-nester to retiree. Life-cycle stage is often more useful than age itself: a 38-year-old with two children in elementary school has very different transport needs from a 38-year-old with no children.
• Gender. Clothing, cosmetics and personal-care marketers have long segmented by gender, though many brands now find that needs cut across gender lines more than old assumptions suggested.
• Income. Luxury brands, financial services and travel companies segment by income. So do discount retailers, which serve shoppers who need low prices. Income tells you what people can afford, not what they value: plenty of high-income households shop at discount stores.
• Household composition. Number and ages of children, number of cars and whether the household owns or rents its home all affect large purchases.

Demographics are a convenient starting point, but they are rarely the full story. Two households with the same income, the same number of children and the same zip code can make opposite choices about cars, bikes, food and vacations, because they hold different values and lead different lives.

7.1.3 Psychographic segmentation

Psychographic segmentation divides buyers by lifestyle, personality, values and interests. It answers the question demographics can't: given two similar-looking people, why does one buy and the other doesn't?

Lifestyle brands depend on psychographics. Patagonia's customers share an outdoor and environmental outlook more than a particular age or income. Harley-Davidson has long marketed to riders who value independence and a sense of rebellion, and its brand community reflects that identity at least as much as any demographic profile. In the e-bike market, a household that values sustainability, exercise and time outdoors may treat a cargo bike as an expression of who they are. A household that sees the bike purely as a tool to cut transport costs may feel no such attachment and will compare it with a used car on price alone.

Psychographic segments explain motivation, but they are harder to measure and reach. You can buy a mailing list of households with children; you can't buy a list of "people who want their kids to grow up less car-dependent."

7.1.4 Behavioral segmentation

Behavioral segmentation divides buyers by their knowledge of, attitude toward, use of or response to a product. Many marketers consider behavioral variables the best starting point, because they are closest to the purchase itself.

• Occasions. When do people need or use the product? Florists segment by holidays and life events; orange juice marketers have long tried to move the drink beyond breakfast. An e-bike can serve a commuting occasion, a school-run occasion, an errand occasion or a weekend-leisure occasion, and the same person may have several.
• Benefits sought. What does the buyer want the product to do for them? Toothpaste buyers may want whitening, cavity protection, sensitivity relief or fresh breath. Benefit segmentation often produces the most actionable segments, because each benefit points directly to a product feature and a message.
• User status. Nonusers, first-time users, regular users, former users and potential users each need different treatment. A former car-owner who is trying to live with one car is a different prospect from someone who has never owned a car.
• Usage rate. Light, medium and heavy users. A small group of heavy users often accounts for a large share of a category's volume, and those users may value durability and service more than a light user does.
• Loyalty status. Some customers are devoted to one brand, some split their loyalty among a few, and some switch to whoever has the best offer. Loyal customers tell you what your brand does well; switchers tell you where you're vulnerable.

Lumen has an advantage here. Its free app records rides, battery health and service needs, so for customers who use the app, Lumen can see actual behavior (how often people ride, how far, at what times) rather than relying on what people say they do. That data is valuable for segmentation, and it also raises privacy questions that this chapter returns to later.

7.1.5 Segmenting business markets

Business markets can be segmented with many of the same ideas, plus variables such as industry, company size, location, purchasing approach and the benefits the buyer organization seeks. Lumen's fleet program, for example, might separate universities (which want bikes for staff getting around large campuses) from delivery businesses (which want durability and low downtime). The buying process behind those segments is covered in Chapter 5. Business Markets and Buyer Behavior. This chapter focuses on consumer markets.

7.1.6 Using several bases together

In practice, marketers rarely rely on one base. They combine several to build a richer picture: a multivariable segment such as "parents of young children (demographic) in dense, hilly cities (geographic) who want to drive less (benefit) and who value time outdoors with their kids (psychographic)." The table below shows how each base applies to Lumen's market.

Base	Example variables	What it might reveal for Lumen
Geographic	Region, city size, density, climate, terrain	Where riding is practical year-round, where hills make e-assist valuable, where dealers can provide test rides and service
Demographic	Age, life-cycle stage, income, household size, car ownership	Households with young children; households with one car or considering selling one; students with little storage
Psychographic	Values, lifestyle, interests	Sustainability-minded households; fitness and outdoors enthusiasts; practical cost-cutters
Behavioral	Occasions, benefits sought, usage rate, user status, loyalty	School-run riders vs. weekend errand riders; heavy daily users who need service; former car commuters

Many companies turn a multivariable segment into a persona, a short, realistic portrait of a typical member of the segment, with a name, a daily routine, goals and frustrations. Personas help a team remember that a segment is made of people. They become dangerous when they are invented rather than drawn from data, because a vivid story is easy to believe. You compared AI-generated personas with real customer reviews in Chapter 4. Consumer Behavior; the same caution applies here.

7.1.7 Requirements for effective segmentation

Not every way of dividing a market produces segments you can use. A good segmentation scheme produces segments that are:

Requirement	Question to ask	Example of failure
Measurable	Can we estimate the segment's size, purchasing power and profile?	"People who secretly want to ride more," with no way to count them
Accessible	Can we reach and serve the segment through media and channels we can use?	A segment concentrated in cities where Lumen has no dealers and no delivery
Substantial	Is it large or profitable enough to be worth a tailored program?	A segment of a few hundred households nationwide
Differentiable	Do members respond differently to marketing than members of other segments?	Splitting buyers by eye color
Actionable	Can we design effective programs to attract and serve it?	A segment whose needs Lumen's products can't meet

7.2 Choosing target markets

Segmentation shows the opportunities. Targeting chooses among them. A company can't serve every segment well, and trying to usually means serving none of them well.

7.2.1 Evaluating segments

To decide which segments to serve, a company evaluates each one on three broad factors.

Segment size and growth. Is the segment large enough today, and is it growing? "Right size" depends on the company. A large company may ignore a segment that would be an excellent business for a small one. For Lumen, a segment of a few hundred thousand households nationally could be very attractive; for a global automaker, the same segment would be a rounding error. Growth matters as much as size, because a growing segment lets a company win customers who haven't yet chosen a brand.

Segment structural attractiveness. Even a large, growing segment can be a poor place to compete. Marketers look at the number and strength of competitors, the threat of new entrants, the availability of substitutes, and the power of buyers and suppliers. A segment full of low-cost online e-bike brands selling under $1,000 is structurally harsh for a premium brand: buyers there compare mainly on price, new entrants appear constantly, and a used bike or bike-share pass is always a substitute. A segment where buyers value safety, service and fit for carrying children may be less crowded and less price-driven.

Company objectives and resources. A segment can be large, growing and structurally attractive and still be wrong for the company. Does serving it fit the company's mission and long-term goals? Does the company have the skills, products, channels and money to win there? Lumen's mission is to "make the everyday trip cleaner, cheaper and more fun than driving," which fits families trying to replace car trips far better than it fits, say, mountain-bike racers. Lumen's resources are also real constraints: a small marketing team, a regional dealer network and a limited budget.

A simple scoring table can make this evaluation explicit. The scores below are illustrative judgments, not data; the point is that writing them down forces a team to argue about each one.

Segment (illustrative)	Size and growth	Structural attractiveness	Fit with Lumen's mission and resources	Overall
Urban families replacing a second car	Medium, growing	Fairly attractive: few direct rivals with strong service	Strong: matches mission and Cargo product	High
Suburban families wanting weekend fun	Large, steady	Mixed: many substitutes, including regular bikes	Moderate: Cargo works, but riding is occasional	Medium
Budget-first commuters	Large, flat	Harsh: low-cost online rivals	Weak: Lumen can't win on price alone	Low
Students in apartments	Medium, steady	Moderate	Good for the Fold, not for the Cargo	Medium (for the Fold)

7.2.2 Targeting strategies

Once it has evaluated the segments, a company chooses how broadly or narrowly to target. The options form a range from very broad to very narrow.

Mass marketing focuses on what buyers have in common. It can deliver economies of scale in production and advertising. It works best when needs really are similar and the product is hard to differentiate, as with many basic commodities. Its weakness is that it is hard to build one product or message that satisfies everyone, and competitors that focus on particular segments can beat a mass marketer in each of them.

Marriott International is a classic example. It operates a range of hotel brands, from luxury properties such as The Ritz-Carlton to midscale brands such as Courtyard and Fairfield, each designed for travelers with different budgets and expectations. Toyota sells most of its cars under the Toyota name and its luxury cars under Lexus. Lumen itself practices differentiated marketing: the Commute is designed for urban commuters, the Cargo for parents carrying kids or groceries, and the Fold for students and apartment dwellers with little storage. Differentiated marketing usually generates more total sales than mass marketing, but it also raises costs: separate product designs, separate messages, more inventory and more management attention.

Concentrated marketing lets a company with limited resources build deep knowledge of one group's needs and a strong reputation among them. Lululemon began by focusing on women who practiced yoga before it broadened into other athletic apparel. The risk is concentration itself: if the niche shrinks or a large competitor decides to enter it, the niche marketer has nothing else to fall back on. Many successful companies start with a niche and expand, which is roughly what Lumen is considering as it takes the Cargo from four Western metro areas to a national market.

Local marketing adapts the offer to a city, neighborhood or even a single store. A retailer may stock different merchandise in a college town than in a retirement community. For Lumen, local marketing could mean different messages in hilly Seattle than in flatter parts of Denver, or dealer events timed to each city's school calendar.

Individual marketing, sometimes called one-to-one marketing, tailors the offer to a single customer. Custom tailors and architects have always done this. Digital tools have made it possible at scale: online stores show different recommendations to each visitor, and apps send different reminders to each user. Individual marketing is where AI has had its largest effect, as you'll see later in this chapter.

Strategy	Who it targets	Main advantage	Main risk	Example
Undifferentiated	The whole market	Low cost, scale	Weak against focused rivals	Basic commodities
Differentiated	Several segments, separate offers	More total sales, better fit	Higher costs and complexity	Marriott's hotel brands; Lumen's three models
Concentrated	One segment or a few niches	Deep knowledge, strong reputation	Dependence on one segment	Early Lululemon
Micromarketing	Local areas or individuals	Highly relevant offers	Cost; privacy; loss of scale	Personalized online recommendations

Which strategy fits depends on the company's resources (limited resources point toward concentration), how much the product and the market vary (uniform products and uniform tastes suit mass marketing), the product's stage in its life cycle (covered in Chapter 8. Products, Services and Brands), and competitors' strategies: a mass marketer facing rivals that segment can be picked apart one segment at a time.

7.2.3 Socially responsible targeting

Targeting decisions can harm people as well as serve them. Marketing that targets children, people in financial distress or other vulnerable groups with products that may hurt them, such as high-interest loans or unhealthy foods, attracts criticism and, increasingly, regulation. The question isn't only who is targeted but whether the targeting serves their interests. Targeting parents with a safety-certified cargo bike that helps them carry their children is easy to defend. Targeting the same parents with fear-based messages about other brands' battery fires, using exaggerated claims, would not be. Chapter 13. Ethics, Responsible AI and Sustainability takes up these questions in more depth.

7.3 Differentiation and positioning

After a company has chosen whom to serve, it must decide how it will compete for those customers. Customers face many choices, and they don't reconsider every brand each time they buy. Instead, they hold simplified pictures of the brands in a category, and they use those pictures to decide which few options to consider.

A product's position is the way customers define it on its important attributes, compared with competitors. Volvo has been associated with safety for decades. Walmart is associated with low prices; its long-running slogan "Save money. Live better." states the position directly. These positions weren't accidents. They are the result of consistent choices about product design, pricing, distribution and communication over many years.

Positioning happens whether a company plans it or not. If Lumen doesn't decide what it should mean to families, customers will decide for themselves, perhaps as "an expensive bike company from Colorado" or "one of those online e-bike brands." The goal of positioning is to make the place in customers' minds a deliberate choice.

7.3.1 Differentiation: finding possible advantages

Positioning starts with differentiation: finding ways the offer can be different from, and better than, competitors' offers for the target market.

A company can differentiate along several dimensions:

• Product. Features, performance, design, durability, reliability or style. Lumen's batteries carry safety certification, a meaningful product difference at a time when news stories about e-bike battery fires are making buyers nervous.
• Services. Delivery, installation, repair, training and support. Lumen's dealers handle test rides and service, and Lumen Care offers an extended warranty and service plan for $149 a year.
• Channels. Coverage, expertise and convenience of distribution. A buyer who can test-ride a Cargo with their kids at a local shop experiences something an online-only brand can't easily match.
• People. Better-trained and more helpful staff. Dealers who know how to fit two child seats and explain battery care are part of the product.
• Image. The brand's identity and associations. Image takes years to build and can't be bought quickly.

Not every difference is worth promoting. A difference is worth establishing when it is:

Criterion	Meaning
Important	It delivers a benefit target buyers value highly
Distinctive	Competitors don't offer it, or the company offers it in a clearly different way
Superior	It is better than other ways customers could get the same benefit
Communicable	Buyers can understand it and see it
Preemptive	Competitors can't easily copy it
Affordable	Target buyers can afford to pay for it
Profitable	The company can offer it at a profit

Battery safety certification, for example, is important to parents, communicable and visible, but how distinctive and preemptive is it? If established bicycle brands also certify their batteries, safety becomes a point of parity rather than a point of difference.

A strong position usually combines a few credible points of difference with enough points of parity that customers don't rule the brand out. A cargo e-bike that offered a brilliant local service network but couldn't safely carry two children would never make a parent's shortlist.

7.3.2 Perceptual maps

A perceptual map (also called a positioning map) is a simple chart that shows how customers see competing brands on two important dimensions. Each axis represents an attribute customers care about, and each brand is plotted where customers place it.

Imagine a map of the family cargo-bike market with price on the horizontal axis (low to high) and support and peace of mind (service, warranty, safety assurance) on the vertical axis. The table below describes where different kinds of competitors might sit. The placements are illustrative, a hypothesis for Lumen's team to test with research, not survey results.

Competitor type	Price	Support and peace of mind	Illustrative placement on the map
Low-cost online e-bike brands	Low	Low: shipped in a box, limited service	Bottom left
Established bicycle brands' cargo e-bikes	High	High: large dealer networks	Top right
Premium specialist cargo-bike makers	Very high	Medium to high	Far right, upper middle
Used cars and second cars	Varies	Familiar, but ongoing costs	Off the bike map: a substitute
Lumen Cargo ($2,999)	Upper middle	High where dealers exist; lower elsewhere	Upper middle, if Lumen can extend service nationally

Maps like this show where the market is crowded, how a brand is actually perceived (which may differ from what its managers believe), and which way a brand must move to reach a desired position. Notice what the Lumen row says: Lumen's strong position on support depends on its dealer network. In cities without Lumen dealers, a national expansion could leave the Cargo looking like an expensive online brand, the worst position on this map.

Perceptual maps should be built from customer data, usually surveys in which customers rate brands on attributes. A map built from managers' opinions is a hypothesis, which is fine as long as everyone treats it as one. The research methods for testing it are covered in Chapter 6. Marketing Research and Analytics.

7.3.3 Choosing a positioning strategy

One useful way to think about position is in terms of the value a brand offers: the benefits it delivers relative to its price. Marketers often describe five winning value positions.

Value position	Benefits vs. price	Example
More for more	Premium benefits at a premium price	Luxury brands such as Rolex
More for the same	Comparable price, more benefits	Lexus, which was introduced as a luxury car offering more than European rivals at a similar or lower price
The same for less	Comparable benefits, lower price	Warehouse clubs and discount retailers
Less for much less	Fewer benefits, much lower price	No-frills airlines and discount grocers such as Aldi
More for less	More benefits at a lower price	Hard to sustain, because more benefits usually cost more to deliver

Lumen can't credibly position the Cargo as "less for much less" against online brands selling under $1,000, and it probably can't sustain "more for less." The realistic choices are more for more (a premium, well-supported family bike) or more for the same (comparable to established bicycle brands on price, but better on something families care about). Which one is right depends on how Lumen's costs and pricing compare, a question taken up in Chapter 9. Pricing.

Companies also position by:

• Attribute or benefit, as Volvo does with safety.
• Use or application, such as "the bike for the school run."
• User, as when a product is positioned for a particular type of person.
• Against a competitor. Avis's long-running "We try harder" campaign turned its position as number two in car rentals into a reason to choose it.
• Away from a category. 7UP positioned itself as "the Uncola," defining itself by what it wasn't.

7.3.4 Writing a positioning statement

A company's chosen position is usually summarized in a positioning statement, an internal document (not an advertising slogan) that guides product, pricing, channel and communication decisions.

A common format is:

To (target segment and need), (brand) is the (frame of reference: the category or concept) that (point of difference), because (reason to believe).

Each part does a job. The target segment and need keeps the statement focused on a particular group. The frame of reference tells customers what kind of thing the brand is, and therefore which alternatives to compare it with. The point of difference is the reason to choose this brand over those alternatives. The reason to believe is the evidence that makes the claim credible.

Here's an example for the Lumen Commute:

To urban commuters who want to get to work without fighting traffic or paying for parking, the Lumen Commute is the e-bike that turns the daily commute into the easiest part of the day, because it is built for daily riding, carries safety-certified batteries, and is backed by local dealer service and an app that warns you before a problem strands you.

You can test any positioning statement with a few questions:

• Is the target specific? "Everyone" or "people who like bikes" is not a target.
• Does the frame of reference set up the right comparison? Calling the Commute "an e-bike" invites comparison with low-cost online e-bikes. Calling it "a replacement for the car commute" invites comparison with driving and parking costs, where it may look better.
• Is the point of difference something customers value and competitors don't offer equally? Check it against the criteria in the differentiation table.
• Is the reason to believe true and provable? If you can't back it up, it doesn't belong in the statement.
• Would it change decisions? A good positioning statement tells the product team what to build, the channel team where to sell and the content team what to say.

7.3.5 Delivering the position

The positioning statement says where the brand sits in the customer's mind relative to competitors. The value proposition you met in Chapter 1. Marketing in an AI Era spells out what the customer gets and why it is worth the price. Every element of the marketing mix must then support both. A "premium, well-supported" position falls apart if the bike is sold at a deep discount, shipped in a box with no assembly help, or advertised in the tone of a bargain brand.

Positions take years to build, so companies change them carefully. Sometimes a brand must reposition because its market has changed. Old Spice, long seen as a brand for older men, repositioned toward younger men with a humorous campaign beginning around 2010. Repositioning is expensive and slow, because the old position lives in customers' memories, which is a good reason to choose a position carefully in the first place.

7.4 AI-driven segmentation and personalization

Segmentation, targeting and positioning were developed when marketers worked with surveys, census data and a few dozen variables. Today, many companies hold far more data: purchase histories, website clicks, app usage, location, service records and customer messages. AI and machine-learning tools can search that data for patterns faster and at a finer grain than any analyst could by hand. This has changed how segmentation is done, and it has made very fine targeting possible. It has also created new ways to get things wrong.

7.4.1 Clustering: letting the data suggest segments

The traditional approach to segmentation is a priori: a marketer chooses the bases first ("let's segment by age and city") and then sorts customers into the groups. Cluster analysis reverses the direction. It starts from the data and looks for groups of customers who are similar to one another on many variables at once.

One common method, k-means clustering, works roughly like this. The analyst chooses how many clusters to look for. The algorithm places that many starting points in the data, assigns each customer to the nearest point, moves each point to the center of its assigned customers, and repeats until the groups stop changing. The result is a set of clusters and a profile of the average customer in each.

Clustering is genuinely useful. It can consider dozens of variables together, it can reveal groups that no one thought to look for, and it can be rerun as new data arrives. A bike company might discover, for instance, that some of the buyers of a model designed for commuters are older riders who use it mainly for weekend recreation, a group the product team never designed for.

But clustering has limits that marketers need to understand:

• It always finds clusters. Ask an algorithm for five clusters and you'll get five, even if the data have no meaningful structure. The groups are a result of the method as much as the market.
• The analyst's choices shape the answer. Which variables go in, how they are scaled and how many clusters are requested all change the result.
• Clusters are statistical, not strategic. A cluster may be real but fail every requirement for effective segmentation: too small to matter, impossible to reach or no different in what it would buy.
• Small or biased data produce fragile clusters. A few unusual customers can create a "segment." Data that include only current customers say nothing about people who chose a competitor.

Generative AI assistants add another layer. Paste a table into a chat and ask for segments, and the assistant will describe groups and give them catchy names. But a chat assistant may not be running a real clustering procedure at all; it may be reading the table and writing a plausible story about it. Its segment names ("Eco-Warrior Parents") can make a weak grouping sound more solid than it is. Treat AI-proposed segments as hypotheses and check them against the data and against the requirements for effective segmentation.

7.4.2 Micro-segmentation, segments of one and personalization

As data and computing power have grown, segments have gotten smaller. Micro-segmentation divides a market into many small, precisely defined groups. Taken to its limit, every customer becomes a segment of one.

Personalization is now routine in digital marketing. Streaming services such as Netflix and Spotify recommend content based on each user's history. Online retailers show different products and offers to different visitors. Email programs change their content by recipient. For Lumen, personalization might mean app messages that suggest a child seat to a Cargo owner who starts riding at school drop-off times, or a Lumen Care reminder timed to a battery's health.

Done well, personalization makes marketing more useful to customers: fewer irrelevant messages, offers that fit. Done badly, it raises several problems.

Privacy. Personalization depends on personal data, and customers don't always know what is collected or how it is used. Location and ride data are especially sensitive, because they can reveal where someone lives, where their children go to school and when the house is empty. Even when data use is legal, it can feel intrusive. Marketers sometimes describe the line between helpful and intrusive as the "creepiness" threshold, and customers who feel watched may stop trusting the brand. Data-protection laws, including the European Union's GDPR and a growing number of U.S. state privacy laws, set rules for consent and use; these are covered in Chapter 13. Ethics, Responsible AI and Sustainability.

Fairness and discrimination. Algorithms that decide who sees an offer can exclude groups of people, even when no one intended it. A model that never uses race or gender directly can still use proxies, such as zip code, that are closely linked to them. In regulated areas such as housing, credit and employment, discriminatory ad delivery is illegal, and major advertising platforms have faced legal action and changed their systems as a result. Outside regulated areas, the effects can still be unfair: some customers may consistently see worse offers or higher prices, which raises questions you'll meet again in Chapter 9. Pricing.

Manipulation. Detailed knowledge of a person's habits and moods can be used to push them toward choices that serve the company more than the customer. Systems that simply maximize clicks or conversions can drift toward this without anyone deciding to.

Loss of the big picture. A company that personalizes everything can lose its positioning. If every customer sees a different message, what does the brand stand for? Strong brands keep a consistent core position and personalize around it.

7.4.3 What stays a human judgment

AI changes the mechanics of STP more than its logic. It can sort customers faster, at finer grain and on more variables. It can draft personas and positioning statements in seconds. It can deliver individual offers to millions of people. But the strategic questions remain human judgments:

• Which segments matter? Choosing a target is a decision about the company's mission, resources and competitive advantage. An algorithm can rank segments by predicted revenue; it can't decide what kind of company you want to be.
• Is the segment meaningful? Someone has to check that clusters are real, reachable and actionable.
• What data is acceptable to use? The book's protect data rule applies with special force here. Customer records don't belong in a public AI tool, and some data shouldn't be used for targeting at all.
• What should the brand stand for? Positioning is a long-term commitment that has to be true to what the company can deliver.

This is the own the decision rule from Chapter 1. Marketing in an AI Era in practice. AI can propose segments; you decide which ones to serve, and you're responsible for how they're treated.

7.5 Lumen Cycles: from "families" to a target market

Back at the whiteboard, Dana's team works through the chapter's steps for the Cargo's national expansion.

Segmenting. Marcus's purchase and app data suggest that Cargo buyers differ most on behavior: how often they ride and what job the bike does for them. Some use the Cargo as a car replacement for school runs and errands nearly every day. Others use it occasionally for errands and weekend outings. Demographics matter (almost all heavy users have children under ten), geography matters (heavy users cluster in denser neighborhoods with schools nearby), and psychographics matter (many describe wanting to drive less). A first segmentation of the national family market might look like this:

Possible segment	Profile	Main benefit sought	Notes
Car-light urban families	Parents of young children in dense city neighborhoods, often with one car	Replace car trips for school and errands	Heavy riders; value safety, weather gear and service
Suburban weekend families	Parents in suburbs with two cars	Family fun and exercise	Ride occasionally; more price-sensitive; compare with regular bikes
Practical haulers	Adults without children who carry gear, groceries or pets	Cargo capacity without a car	Smaller group; less interested in child seats
Fleet buyers	Universities, campuses, delivery businesses	Durable hauling, low downtime	Covered by the fleet program, not the consumer plan

Targeting. The team scores each segment on size and growth, structural attractiveness, and fit with Lumen's mission and resources. Car-light urban families fit the mission ("cleaner, cheaper and more fun than driving") most directly and value the things Lumen does best: safety-certified batteries, service and a bike built for carrying kids. But the national expansion raises an accessibility problem. Lumen's dealers, who provide test rides and service, are concentrated in the Mountain West and Pacific Northwest. A national target only makes sense if Lumen can serve families where it has no dealers, perhaps by recruiting new dealers in selected cities or by offering mobile service. That is a channel question for Chapter 10. Channels, Retail and the Customer Journey, but it shapes the targeting decision now. One reasonable answer is a concentrated strategy: target car-light urban families first, in a short list of dense cities where Lumen can provide test rides and service, and expand from there.

Positioning. On the illustrative perceptual map, the Cargo's best position combines family safety with local support at an upper-middle price, above low-cost online brands and somewhat below premium specialist cargo bikes. The team must check whether that position is distinctive. If established bicycle brands also offer certified batteries and dealer service, Lumen needs a sharper point of difference, perhaps how much easier it makes daily life without a second car, supported by the app, Lumen Care and the experience of current Cargo families.

Meanwhile, the Commute faces a different positioning problem. Low-cost online rivals are pulling price-focused commuters away. Lumen can't win "less for much less," so the Commute's position must emphasize what those rivals lack, such as reliability, safety certification and local service, and Lumen must accept that it won't win every price-focused buyer.

Your marketing plan. Draft the Segments, target market and positioning statement section of your plan (for Lumen, or for the company your instructor approved). Include:

1. Segmentation. Identify three to five segments in the market, using at least two bases (for example, behavioral and demographic). Describe each segment's profile and the main benefit it seeks.
2. Segment evaluation. Score each segment on size and growth, structural attractiveness, and fit with the company's objectives and resources. State which scores are supported by evidence and which are your judgment.
3. Target market and strategy. Choose your target segment or segments and name the targeting strategy (undifferentiated, differentiated, concentrated or micromarketing). Explain why, and check your target against the five requirements for effective segmentation.
4. Positioning. Describe the main competitors your target customers consider, sketch or describe a perceptual map on two dimensions those customers care about, and identify your points of parity and points of difference.
5. Positioning statement. Write a positioning statement in the chapter's format, and show how it connects to the value proposition you drafted earlier.
6. AI use. If you used AI for any part of this section, note what you used it for and how you checked the result, following the disclose it rule.

7.6 Try it with AI

In this exercise, you'll give an AI assistant a small customer dataset, ask it to propose segments, and then judge whether its segments are meaningful, reachable and actionable using this chapter's criteria.

The table below is a fictional sample of 14 Lumen customers, made up for this exercise. Because it's fictional, you can paste it into an AI tool. In a real job, you would never paste actual customer records into a public AI tool; that would break the protect data rule. You'd use an approved internal tool, or remove anything that identifies a person before analysis.

customer_id	model	age	kids_under_10	city	purchase_channel	main_use	rides_per_week	miles_per_week	lumen_care	accessories	top_reason_for_buying
C01	Cargo	38	2	Denver	Dealer	School run and errands	9	38	Yes	Child seat, rain cover	Replace our second car
C02	Commute	29	0	Seattle	Website	Commute to work	8	60	No	Lock	Faster than the bus
C03	Fold	21	0	Salt Lake City	Website	Campus trips	10	25	No	Lock	Fits in my apartment
C04	Cargo	41	1	Portland	Dealer	Groceries and errands	5	22	Yes	Rack, panniers	Fewer car trips
C05	Commute	52	0	Denver	Dealer	Weekend recreation	3	35	Yes	None	Fitness and fun
C06	Cargo	35	2	Seattle	Website	School run and errands	10	40	Yes	Two child seats	Kids love it
C07	Fold	33	0	Portland	Website	Commute (train plus bike)	10	30	No	None	Fits on the train
C08	Commute	31	0	Denver	Website	Commute to work	9	70	No	Lock, lights	Save on parking
C09	Cargo	44	3	Salt Lake City	Dealer	Groceries and errands	4	15	No	Child seat	Fewer car trips
C10	Commute	58	0	Seattle	Dealer	Weekend recreation	2	30	Yes	Rack	Keep up with friends on rides
C11	Fold	20	0	Denver	Website	Campus trips	8	18	No	Lock	Cheap to run
C12	Cargo	37	2	Denver	Website	School run and errands	8	30	Yes	Child seat, rain cover	Replace our second car
C13	Commute	27	0	Portland	Website	Commute to work	7	55	No	Lock	Cheaper than driving
C14	Commute	63	0	Salt Lake City	Dealer	Weekend recreation	3	40	Yes	None	Hills are easier

Start a new conversation, paste the table, and ask for segments. Asking the AI to show its reasoning makes its output easier to check.

Now check the reply. Don't take the segment names or descriptions on trust.

1. Check the assignments against the data. For each segment, look up every customer_id in the table. Does each customer really match the profile the AI described? Did the AI leave anyone out, put anyone in two segments, or describe a customer with details that aren't in the table? Any invented detail is a reason to distrust the rest.
2. Check the bases. Did the AI simply group customers by model (Cargo, Commute, Fold)? That's a product grouping, not a segmentation of needs. Look for groupings that cut across models or split a model, such as the older Commute buyers who ride for recreation rather than commuting.
3. Apply the five requirements. For each proposed segment, decide whether it is measurable, accessible, substantial, differentiable and actionable. Fourteen customers can't tell you whether a segment is substantial in the national market; note what data you would need to find out.
4. Judge the confidence statement. A good reply admits that 14 rows are far too few to establish segments, and treats them as hypotheses. If the AI sounds certain, discount its certainty.

Next, push the AI from segmentation to targeting. This tests whether it can apply the chapter's evaluation criteria rather than just describe groups.

As you read the second reply, look for these problems:

• Invented market facts. If the AI states national market sizes, growth rates or survey findings, it didn't get them from your table. Apply the verify it rule: either find a real source for each figure or treat it as unsupported.
• Ignored constraints. Did the AI account for Lumen's regional dealer network when it recommended a national target? A recommendation that ignores accessibility fails the chapter's test.
• Missing judgment. Compare the AI's recommendation with your own scoring. Where you disagree, decide who has the better argument and why.

Finally, write a short note (one paragraph) for your marketing plan's AI-use appendix: what you asked, what you kept, what you rejected, and how you checked it. That's the disclose it rule in practice.

7.7 Summary

Market segmentation divides buyers into groups with different needs or responses, using geographic, demographic, psychographic and behavioral bases, usually in combination; useful segments are measurable, accessible, substantial, differentiable and actionable. Targeting evaluates segments on size and growth, structural attractiveness and fit with the company, then chooses a strategy from undifferentiated marketing to micromarketing. Positioning decides what place the brand should hold in target customers' minds, using points of parity and difference, perceptual maps and value positions, and captures the choice in a positioning statement that guides the whole marketing mix.

AI has made segmentation faster and finer, through clustering and personalization. But clusters appear whether or not they mean anything, chat assistants can dress weak groupings in persuasive names, and fine-grained targeting raises privacy, fairness and manipulation concerns. Which customers to serve, what data to use and what the brand stands for remain human judgments.

7.8 Review Questions

1. Define market segmentation and market segment. Why is "respond in a similar way to marketing" essential to the definition of a segment?
2. Describe the four main bases for segmenting consumer markets and give an example of each for a smartphone brand.
3. List the five requirements for effective segmentation. Choose one segment you belong to (as a customer of any product) and evaluate it against all five.
4. A regional craft brewery is deciding between concentrated and differentiated marketing. What factors should influence its choice, and what are the main risks of each strategy?
5. Explain the difference between a point of parity and a point of difference. Why does a brand need both?
6. Sketch a perceptual map for fast-food or fast-casual restaurants near your campus, using two dimensions you think customers care about. Where are the gaps, and would any of them be worth filling? What data would you need to check your map?
7. An AI tool clusters a retailer's customers into seven segments and gives each a catchy name. Describe three checks you would make before using these segments in a marketing plan.
8. Discussion: Lumen could use app ride data to send personalized offers to Cargo owners based on when and where they ride. Where would you draw the line between helpful personalization and intrusive targeting? Explain your reasoning with reference to the book's rules for AI use.